[PATCH 2/3] opp: Add devres wrapper for dev_pm_opp_set_prop_name

Frank Lee tiny.windzz at gmail.com
Fri Oct 30 11:19:00 UTC 2020


On Wed, Oct 28, 2020 at 10:46 PM Viresh Kumar <viresh.kumar at linaro.org> wrote:
>
> On 28-10-20, 19:02, Frank Lee wrote:
> > On Wed, Oct 28, 2020 at 6:29 PM Viresh Kumar <viresh.kumar at linaro.org> wrote:
> > >
> > > On 12-10-20, 21:55, Frank Lee wrote:
> > > > From: Yangtao Li <tiny.windzz at gmail.com>
> > > >
> > > > Add devres wrapper for dev_pm_opp_set_prop_name() to simplify driver
> > > > code.
> > > >
> > > > Signed-off-by: Yangtao Li <tiny.windzz at gmail.com>
> > > > Signed-off-by: Yangtao Li <frank at allwinnertech.com>
> > > > ---
> > > >  drivers/opp/core.c     | 39 +++++++++++++++++++++++++++++++++++++++
> > > >  include/linux/pm_opp.h |  6 ++++++
> > > >  2 files changed, 45 insertions(+)
> > >
> > > On a second thought I am looking at dropping this one as you haven't
> > > added any users yet and I am afraid it will stay unused.
> >
> > Now it looks like that dev_pm_opp_set_prop_name() is used relatively less.
> > Maybe we can wait until a caller, and then pick up the patch.
>
> I am even wondering if we should be adding any of the devm_* helpers
> for now to be honest. Even for the other one we have only one user.
> Them major user of the OPP core is the CPU subsystem and it is never
> going to use these devm_* helpers as the CPU device doesn't get bound
> to a driver, it is rather a fake platform device which gets the
> cpufreq drivers probed. So the only users of these devm_* helpers is
> going to be non-CPU devices. Considering that we have only one user
> right now, it may be better to just fix it instead of adding any of
> the devm_* helpers.

GPU is also a relatively large number of opp consumers.
Most of the time, the dev_pm_opp_set_* functions will only be set once.
If don't need the driver to dynamically manage and release the opp, it
may be OK?

Yangtao


More information about the dri-devel mailing list